pp. 1847-1869
S&M1903 Research Paper of Special Issue https://doi.org/10.18494/SAM.2019.2308 Published: June 7, 2019 Recognizing Falls, Daily Activities, and Health Monitoring by Smart Devices [PDF] Sittichai Sukreep, Khalid Elgazzar, Cheehung Henry Chu, Chakarida Nukoolkit, and Pornchai Mongkolnam (Received January 21, 2019; Accepted March 25, 2019) Keywords: health monitoring, smart devices, IoT, daily activity, fall recognition, data mining, classification
One of the biggest challenges in ageing societies is to improve life, health, safety, and support of the elderly population in their daily life. Currently, the number of elderly people living alone is increasing every year. Living alone allows more freedom but raises the risk of serious injuries or fatal accidents. Falls are the key cause of significant health problems, especially for an elderly person who lives alone. Moreover, vital signs such as heart rate, balancing activities, and environmental context are crucial in relation to the user’s condition. To assist people living alone and improve their health quality, we firmly believe that the advances in Smart Devices, Smart Environment, and Internet of Things paradigms are very helpful for developing a fall and activity recognition system. We propose a system using an unobtrusive device consisting of a smartwatch and a smartphone to identify falls and thirteen daily activities (e.g., walking, running, typing, and waving the hand). The events leading to a fall, the speed of falling down, the heart rate while doing an activity, and the time passed since the fall are important data that we store to help a doctor diagnose and rehabilitate a patient. Environment sensors are used to indicate the indices of ambient conditions such as temperature, humidity, brightness, and motion detected. Suitable machine learning techniques are used for daily activity recognition, and the processing time for classification was compared on the basis of a smartwatch and an Amazon Web Services (AWS) cloud server. Threshold-based health risk analysis models are utilized for abnormal activity recognition and heart rate and heat index (temperature and humidity) determination. The system issues different types of notifications such as warning messages, sounding alarms, and phone calls to related persons such as family members, caregivers, or doctors. Various easy-to-understand visualizations are presented to track and monitor the subjects in real time, including heart rate, daily activity summary, health risk status, and environmental information.
Corresponding author: Sittichai SukreepThis work is licensed under a Creative Commons Attribution 4.0 International License. Cite this article Sittichai Sukreep, Khalid Elgazzar, Cheehung Henry Chu, Chakarida Nukoolkit, and Pornchai Mongkolnam, Recognizing Falls, Daily Activities, and Health Monitoring by Smart Devices, Sens. Mater., Vol. 31, No. 6, 2019, p. 1847-1869. |